This investigated the impacts of ratings of larger groups of liked and disliked people would have on the similarity levels. They predicted that by asking participants to think of many celebrities, they would be more likely to pick a diverse set of people; thus, influencing how similar they were all to each other. They asked participants to think of 12 celebrities they liked and disliked; however, only 4 of each were randomly selected for the participants to make comparison ratings about (e.g. Celebrity 1 is more sociable than Celebrity 2).